Abstract
Brain-machine interfaces (BMIs) for reaching have enjoyed continued performance improvements. Yet there remains significant need for BMIs that control other types of movement, including ego-motion through the world. Based on recent scientific findings, the class of decode algorithms employed by reach-based BMIs seems unlikely to generalize well. To examine this, we developed an ego-motion BMI based on cortical activity as monkeys cycled a hand-held pedal to progress along a virtual track. Unlike reaching, strong correlations between neural activity and kinematics were not present during cycling. We thus employed an opportunistic decode strategy that abandoned any notion of inverting encoding, and instead identified and leveraged dominant features of the population response. Online BMI-control success rates approached those during manual control. We argue that, in retrospect, reach-based BMIs succeeded by implicitly using opportunistic strategies, and that explicitly embracing that approach will be essential for expanding the range of BMI-controllable movements.
Competing Interest Statement
The authors have declared no competing interest.
Footnotes
Updated Abstract, Introduction, and Discussion sections.